Model be worthier than it's previous

vision
1

I’m working on 3d landmark detection model
the training process works very well for el first 2 epochs and at the middle of the epoch 3 the loss starts to increase in an insane way and model results get much worth than it’s previous
what could be the reason of this

2

Is the training loss or the validation loss going up?
Could you post the loss curves for both losses?

3

for frist two epochs
trainloss inf -> 0.1 -> 0.03
validationloss inf->0.8->0.021
when this issue happens in the first 30% of epoch 3 i stop train to check the results in real images
the results was very bad compared to the previous model